RAS Mutations Beyond KRAS Exon 2: A Review and Discussion of Clinical Trial Data

Opinion Statement

The addition of targeted therapy to a 5-FU chemotherapy backbone is now a standard of care in metastatic colorectal cancer. Epidermal growth factor receptor (EGFR) inhibitors have been demonstrated to improve progression-free survival (PFS) and overall survival (OS) in the first line for patients with tumors that do not harbor KRAS exon 2 mutations.

Eligibility criteria for most clinical trials involving EGFR inhibitors in recent years have
used the absence of KRAS exon 2 mutation as the sole criteria for entry, as this specific
mutation has been consistently shown to be predictive of a poor response to EGFR
inhibitors. However, expanded analyses of first-line metastatic trials reveal that other
RAS mutations, such as other KRAS mutations in exons 3 and 4, along with NRAS
mutations, are predictive of poor responses to EGFR inhibitors as well. Testing for a full
panel of these RAS mutations should be done prior to initiating treatment with an EGFR
inhibitor. Further clinical trials are required to determine the predictive impact of each of
these individual mutations. To date, they have been analyzed in the aggregate. The
addition of targeted therapy, bevacizumab or an EGFR inhibitor, to a chemotherapy

Introduction

Colon cancer is the second leading cause of cancer death in America, with over 50,000 deaths projected in the USA in 2014 [1]. Outcomes in advanced colorectal can- cer have improved with the addition of targeted therapy to the chemotherapy 5-fluorouracil (5-FU) backbones. Epidermal growth factor receptor (EGFR) inhibitors, such as cetuximab and panitumumab, have become an essential part of the armamentarium of medical oncol- ogists that treat colorectal cancer. Cetuximab initially was used in refractory patients, after demonstrating su- perior overall survival (OS) and progression-free surviv- al (PFS), compared to best supportive care [2]. In the past decade prior to this publication, clinical trials have established the effectiveness of cetuximab in combination with irinotecan [3], 5-FU, 5-FU plus irinotecan (FOLFIRI), and 5-FU plus oxaliplatin (FOLFOX) [3, 4, 5••].

The initial interest in EGFR inhibitors developed with discovery of regulation of the EGFR gene in 70 to 80 % of colorectal cancer cases [6]. Binding of ligand, mainly epidermal growth factor (EGF) and TGF-α [6], to the extracellular domain of EGFR in colorectal cells incites autophosphorylation of the internal tyrosine ki- nase domain. This leads to downstream signaling through multiple pathways, including RAS, as depicted in Fig. 1 [7]. The net result is cell proliferation, angio- genesis, and other cancer-promoting activities [8].

Naturally, the initial pursuit of a predictive biomark- er was directed toward EGFR expression. Interestingly, phase II studies by Saltz et al. and Cunningham et al. published in 2004 showed no correlation between EGFR expression and response to cetuximab [9, 3]. In the former study, EGFR expression was measured both as a percentage of EGFR-positive cells and in terms of staining intensity, with neither correlating with response.

KRAS, an important component of the EGFR down- stream signaling cascade, emerged as a predictive bio- marker in 2008. Karapetis et al. published results from an analysis of tumor samples from the Jonker et al. study published 1 year earlier [10]. He found that just over

43 % of the tumors demonstrated KRAS mutations in exon 2. Furthermore, the population with that mutation derived no benefit from cetuximab, while the median survival was nearly doubled in the treatment group among the KRAS wild-type group [10]. Several other studies that retrospectively evaluated KRAS exon 2 mu- tational status in EGFR inhibitor clinical trials corrobo- rated these results [11, 12].
KRAS is a proto-oncogene that, when turned on, leads to downstream events that impact cell division. It encodes a protein that links activation of KRAS to the EGFR pathway. Codons 12 and 13 on exon 2 are most commonly mutated, resulting in constitutive activation of KRAS [13, 14]. The discrepancy in responses to EGFR inhibition is due to a KRAS mutant pathway of carcino- genesis that operates independently of the ligand- activated EGFR pathway. For these tumors, the Ras/MAPK pathway is constitutively activated re- gardless of EGFR expression [8, 15, 16].
Evaluation of KRAS exon 2 prior to initiation of EGFR inhibitors has become a standard of care. Cetuximab, a chimeric monoclonal antibody, and panitumumab, which is a humanized monoclonal anti- body, both demonstrated effectiveness in the colorectal cancer population with KRAS wild-type tumors [3, 17, 18]. Mutations in exon 2 of KRAS were found to be excellent at predicting who would not respond to cetuximab [15, 19, 20]. However, the positive predictive value for response to EGFR-targeted agents with KRAS wild-type status is 40–60 %, suggesting the need for the development of additional predictive factors [15].

More recent clinical trials evaluating the use of EGFR inhibitors excluded patients with KRAS exon 2 muta- tions, but included patients with KRAS mutations at loci other than exon 2 [4, 5••] (heretofore called Bother RAS mutation^). With the availability of mass genotyping assays that have made it easier to analyze less common mutations, other RAS mutations were than analyzed for their predictive value (Fig. 2). The following is a review and analysis of the major clinical trials that evaluated outcomes in RAS mutations outside of exon 2 and a discussion of the clinical and research implications of Bother RAS mutations.^ We searched for prospective trials of first-line therapy with an EGFR inhibitor that involved greater than 30 patients with other RAS mu- tations that were treated with an EGFR inhibitor. Here, we focus on six studies; PRIME [21], PEAK [22], CRYSTAL [23], OPUS [24], FIRE-3 [4, 25], and CALGB/SWOG 80408 [5••, 26]. Though the COIN trial [27] did include an expanded analysis, this is not described in detail, given that KRAS codon 4, the most commonly seen other RAS mutation, was not evaluated in their study.

PRIME

The PRIME study was a comparison of FOLFOX4 with and without
panitumumab. The study initially analyzed all patients without an exon 2 mutation. In this subgroup, the PFS in combination was 9.6 vs. 8.0 months in and OS was 23.8 months in favor of the panitumumab-FOLFOX4 group.

Their analysis of other RAS mutations was a prospective-retrospective bio- marker analysis of RAS mutations, designed to analyze the treatment effect of panitumumab. Among the 108 patients with RAS mutations, they found that other RAS mutations predicted for a lack of response to panitumumab. In the other RAS mutation subgroup, the PFS was 8.0 for FOLFOX alone, compared to 7.3 for panitumumab-FOLFOX4, and the OS was 17.8 and 17.1, respectively (see Tables 1 and 2). In both categories, FOLFOX alone seemed favorable, but did not reach statistical significance [21, 28].

PEAK

Fig. 2. Schematic demonstrating the incidence of each individual mutation in the studies highlighted. Overall, KRAS exon 4 was most common, occurring in 4.9–9.3 % of patients in these clinical trials.

The PEAK study was a randomized phase II study that randomized KRAS exon 2 WT patients to FOLFOX+panitumumab vs. FOLFOX+bevacizumab. There were 278 patients treated in the trial. There was no significant difference in PFS. As for OS, which was a secondary endpoint, the FOLFOX+panitumumab group was superior (34.2 vs. 24.3 months).

It included a predefined analysis of patients with RAS mutations outside of exon 2. There were a total of 51 patients (23 %) with other RAS mutations that were evaluated. The OS for those in the cohort that was absent of any RAS mutations was 41.3 months.

The exclusion of other RAS mutations increased the difference in PFS between the
panitumumab and bevacizumab groups from 4.2 to 5.6 months. Among the other RAS mutation subgroups, those that received panitumumab had a worse PFS when compared to the group receiving bevacizumab. Somewhat surprisingly, among this small population, the OS was much better for those receiving panitumumab (see Table 2) [22].

CALGB/SWOG 80405

The CALGB/SWOG study was presented during the plenary session at ASCO 2014. It is the largest study that analyzed patients with an extended RAS analysis. A total of 1140 patients with KRAS exon2 wild type were randomized to chemotherapy plus cetuximab vs. chemotherapy plus bevacizumab. They allowed for physician choice regarding the chemotherapy backbone (FOLFOX vs. FOLFIRI). The primary endpoint was OS, and there was While KRAS testing has aided in selection of patients that are likely to benefit from EGFR inhibition, many patients that are exon 2 wild type do not benefit from treatment. A more accurate selection of patients that may benefit from these agents is needed. Patients taking EGFR inhibitors suffer from side effects such as dermatologic toxicity that can adversely impact quality of life. The severity of these side effects has compelled some physician/researchers to consider using these agents only in the second line [34] regardless of RAS status. This is despite the proven OS and PFS advantages over chemotherapy alone in first line, and at least equivalent results to chemotherapy+bevacizumab.

The choice of which targeted agent to add to first-line therapy, agents that act upon the VEGF pathway or EGFR inhibitors, may take on greater importance since the results of the ML18147 trial were published. This established the advantage of continuation of bevacizumab on progression over chemotherapy without targeted therapy [35]. This study’s results have impacted decision making in the second line, making practitioners feel married to bevacizumab once they have initiated it. This highlights that more predictive information is needed to make the best decision about targeted agents being added to the backbone of chemotherapy for first-line therapy.

The studies discussed above confirm that extended analysis of RAS muta- tions beyond exon 2 should now be a standard of care. As confirmed in a meta- analysis by Sorich et al. [36•], the findings in these studies and others [27] demonstrate that other RAS mutations, when analyzed in aggregate, do not benefit from EGFR antibodies. Outcomes in terms of both PFS and OS are no better with use of an EGFR inhibitor in this population (see Tables 1 and 2). Based on the above studies, 14.7–23 % of patients that would previously be considered as good candidates for EGFR inhibition can now be appropriately steered toward other treatment options. This will prevent unnecessary side effects and allow practice of cost-effective medicine (1 month of cetuximab is reported to cost $10,000) [37].

Members of the RAS family play a role in many human tumors including colorectal cancer, myeloid leukemia, and melanoma [13, 38, 39]. They are small GTPases that regulate cell division by alternating between GDP and GTP- bound states [13]. The enzymes encoded by HRAS, KRAS, and NRAS are genotypically similar, being distinguished by their hypervariable region in the carboxy-termini. The enzymes encoded by RAS share the common function of regulating growth and apoptosis [40]. RAS proteins that are mutated have insensitivity to GTPase-activating proteins and are constitutively activated.

These permanently activated RAS proteins cause overactive signaling without stimulus. This leads to uncontrolled cell division.KRAS mutations, such as the less common exon 3 and exon 4 mutations, were not thoroughly investigated when EGFR inhibitors were first studied in clinical trials. Only when means to perform mass genotyping assays became more available were lower frequency mutations such as exons 3 and 4, as well as NRAS mutations, analyzed as predictive markers of response EGFR inhibition [41].
Mutations in KRAS exons 2, 3, and 4 are mutually exclusive, as are KRAS and NRAS mutations. This suggests functional redundancy [42]. There is significant overlap in their function, but some differences have been described. While codons 12 and 13 (exon 2) impair RAS GTPase activity, this is not seen with codon 146 mutations. Rather, it is thought to be an increase in guanine nucleotide exchange that leads to oncogenic transformation [13, 42, 43]. MEK dependence may be a unique feature of exon 4 mutation as well [44, 45].

KRAS and NRAS mutations appear to arise in response to different types of stress. As originally described in melanoma models [46], KRAS is mutated early in the process of oncogenesis, as suggested by the occurrence of these mutations in adenoma formation [45]. NRAS mutations seem to occur once oncogenesis has begun and promote ongoing tumor growth through suppression of apo- ptosis [46].

While they may have some difference in function, it is not clear whether the distinction between KRAS exon 2 mutations and other RAS mutations is im- portant in terms of predictive or prognostic value. A major deficiency in our current knowledge of other RAS mutations is the prognostic implications of each individual mutation.

Currently, there is no rationale for tailoring therapy based on the specific RAS mutations. No analysis of individual mutations has been done, and KRAS and NRAS other mutations are all analyzed in the aggregate. While overlap in function of various RAS proteins makes a meaningful difference seem unlikely, further analysis needs to be done to rule out outliers. Within exon 2, for example, codon 13 has been suggested to be a weaker stimulus of oncogenic transformation than codon 12, which prompted De Roock et al. to evaluate this subgroup’s outcomes when given cetuximab in clinical trials. Among their small subgroup of 32 patients, there did seem to be a slight benefit from EGFR inhibition in the group with codon 13 mutations, suggesting the need for prospective evaluation [47]. The FIRE-3 trial also showed difference outcomes for exons 12 and 13. A codon 13 mutation was associated with a higher response rate, but a shorter PFS and OS. There has been some sugges- tion that the most common Bother^ RAS mutation, KRAS exon 4, correlates with a better prognosis [44].

Conclusion

The trials highlighted above were both comparisons of the addition to EGFR inhibitors to standard chemotherapy and comparisons of bevacizumab to EGFR inhibitors in first-line therapy. In both settings, the extended RAS muta- tion group did not benefit from the use of EGFR inhibition (see Tables 1 and 2). The question of whether adding cetuximab for this population is inferior to bevacizumab is still unclear, though the phase II FIRE-3 trial demonstrated a 6- month difference in PFS favoring bevacizumab. Likewise, the question of whether cetuximab results in inferior outcomes for RAS-mutated patients when compared to chemotherapy alone is unanswered; the PRIME and OPUS studies suggest a worse outcome, while CRYSTAL did not.

Among all of the studies reviewed, KRAS exon 4 was the most common non- exon 2 mutation found. NRAS exon 4 was exceedingly rare, with most of the studies not recognizing any cases (See Fig. 2). HRAS, a mutation found in other malignancies, appears to be rare in colorectal cancer and was not assessed in these studies [48]. The findings varied slightly across the above studies in terms of number and types of other RAS mutations found (Table 2). As described in The Oncologist [49•], investigators sent 10 samples with more rare RAS muta- tions to 131 laboratories and found that genotypic mistakes were made in over 25 % of the centers. In the above studies, techniques for analysis varied both in the amount of tumor analyzed and the method used for detection of RAS mutations. For example, the investigators in the CRYSTAL trial had a patholo- gist estimate tumor content of specimens without micro- or macro-dissection, using an estimate of 5 % as a cutoff for analysis, while the PRIME study macro- dissected tumors with less than 50 % tumor burden. CRYSTAL used the BEAM technique [50] to identify RAS mutations, while bidirectional Sanger sequenc- ing with WAVE-based surveyor kits were used in PRIME.

Another area of future investigation may be exploring outcomes based on the burden of RAS mutation. The CRYSTAL analysis used a cutoff of 5 % mutant to wild-type ratio as the criteria for Bmutated.^ However, they also analyzed 23 patients that had more than 0.1 % RAS mutations detected by their BEAM technique, but less than 5 %. This subgroup did benefit from cetuximab, suggesting that there is a correlation between burden of RAS mutation in neoplastic cells and resistance to EGFR inhibition. Similar findings were re- ported by Laurent-Puig et al., who noted that patients with G1 % mutated KRAS alleles have a similar response to wild-type KRAS patients [51].
Ultimately, the decision to give cetuximab or panitumumab will depend on factors beyond RAS mutations. BRAF mutations, which are mutually exclusive with RAS mutations, in colorectal cancer, are associated with a very poor prognosis, regardless of first-line treatment [52]. Though response to EGFR inhibition has been poor, there is some interest in combined inhibition of EGFR and BRAF [4, 53, 54]. PI3K mutations also appear to be predictive of resistance to EGFR inhibition, further shrinking the pool of good candidates for these agents [55–59]. Some possible positive predictive markers are emerging for further exploration, such as HER-3 and CD73 [60].

In the aggregate, low-frequency KRAS and NRAS mutations appear to be ap- proximately as resistant to EGFR inhibition as KRAS exon 2 mutations.

Extended analyses including KRAS codons 3 and 4, as well as NRAS codons 2, 3, and 4, should be performed in all patients that would be considered for targeted therapy with cetuximab or panitumumab. Based on current knowledge, EGFR inhibitors such as cetuximab or panitumumab should be avoided in patients whose tumors harbor any of these mutations. However,RMC-9805 further analysis is needed to analyze the predictive value of each individual mutation.